According to the ability of limited decomposition directional subband and difficult to suppress noise based on the traditional multi-scale analysis, a multi-focus image fusion method based on Surfacelet transform and composite incentive model is proposed. Original images are decomposed by Surfacelet transform to obtain a number of different frequency band sub-images. A composite incentive model is built based on the characteristics of the low frequency sub-band and high-frequency sub-band coefficients, namely improved-sum-modified-Laplacian and spatial frequency are selected as external stimulus of compound PCNN. Fusion coefficients are preferred by compound PCNN and the results are improved. The experimental results show that grayscale distribution of the fusion image is more dispersed and coherent image texture details are outstanding. The algorithm overcomes the traditional multi-focus image fusion defects, and the objective evaluation indexes show that this method is superior to that of Laplace, Discrete Wavelet Transform (DWT) and PCA traditional image fusion methods.